Researchers at the World Institute of Kimchi examined a kimchi sample using hyperspectral imaging processed by an AI model at their facility in Gwangju, as captured in a photo from December 2023. The credit for the image goes to the World Institute of Kimchi.
Authored by Ko Dong-hwan, the adoption of data-driven artificial intelligence (AI) is transforming the quality assurance process for mass-produced kimchi aimed at global consumers, according to a state-operated kimchi laboratory. This technological innovation signifies a departure from the traditional method of preparing pickled cabbage, which historically relied on instinctual techniques.
The World Institute of Kimchi (WiKim) introduced this cutting-edge technology following a successful six-month collaboration that concluded in December. The partnership involved a local AI developer and a digital education consultancy with the main objectives of establishing a comprehensive dataset for kimchi quality assessments and developing an AI model capable of analyzing the gathered data effectively.
In collaboration with Seoul-based data solution provider Catalonix and edu-tech service developer SLI (Solution Learning Innovation), both headquartered in Seoul, WiKim curated a vast dataset comprising 270,000 entries of RGB color and hyperspectral images. This dataset enabled a detailed evaluation of food quality throughout various stages of kimchi production, including cabbage salting, seasoning mixing, and fermentation processes.
Utilizing this dataset, the AI model can analyze images to determine sweetness, saltiness, and fermentation levels at each stage of kimchi production, crucial factors that significantly influence the overall quality of the dish.
The dataset and AI model work together to provide accurate quality assessments for kimchi manufacturing facilities, covering ingredient quality, mixing processes, and fermentation monitoring. Additionally, the system can grade the kimchi based on different attributes, offering a comprehensive quality evaluation of the final product.
President Chang Hae-choon of WiKim emphasized the efficiency of this technology in expediting the inspection of mass-produced kimchi from production to distribution, maintaining consistently high quality standards compared to traditional methods.
This advancement is set to enhance South Korea’s kimchi industry, which has traditionally lacked standardized quality measures, relying heavily on subjective judgments. With the global popularity of Korean culture driven by K-pop and social media, ensuring consistent quality for kimchi consumers worldwide is crucial.
Chang highlighted the need to address challenges in the kimchi export process, emphasizing the importance of departing from traditional methods to adopt innovative approaches for quality assessment. The AI-driven model for kimchi quality evaluation has overcome past limitations, enhancing production efficiency.
WiKim’s progress under the Ministry of Science and ICT aligns with the institute’s selection by the National Information Society Agency for a national R&D project focused on creating a dataset for AI learning.
